Cationic porphyrins are tunable gatekeepers of the
20S proteasome
†
Anna M. Santoro,aAlessandra Cunsolo,bAlessandro D'Urso,bDiego Sbardella,c Grazia R. Tundo,cChiara Ciaccio,cMassimiliano Coletta,*cDonatella Diana,d Roberto Fattorusso,*eMarco Persico,fAntonio Di Dato,fCaterina Fattorusso,*f Danilo Milardi*aand Roberto Purrello*b
The 20S proteasome is a barrel-shaped enzymatic assembly playing a critical role in proteome maintenance. Access of proteasome substrates to the catalytic chamber is finely regulated through gating mechanisms which involve aromatic and negatively charged residues located at the N-terminal tails ofa subunits. However, despite the importance of gates in regulating proteasome function, up to now very few molecules have been shown to interfere with the equilibrium by which the catalytic channel exchanges between the open and closed states. In this light, and inspired by previous results evidencing the antiproteasome potential of cationic porphyrins, here we combine experimental (enzyme kinetics, UV stopped flow and NMR) and computational (bioinformatic analysis and docking studies) approaches to inspect proteasome inhibition bymeso-tetrakis(4-N-methylpyridyl)-porphyrin (H2T4) and
its two ortho- and meta-isomers. We show that in a first, fast binding event H2T4 accommodates in
a pocket made of negatively charged and aromatic residues present ina1 (Asp10, Phe9), a3 (Tyr5), a5 (Asp9, Tyr8),a6 (Asp7, Tyr6) and a7 (Asp9, Tyr8) subunits thereby stabilizing the closed conformation. A second, slower binding mode involves interaction with the grooves which separate thea- from the b-rings. Of note, the proteasome inhibition byortho- and meta-H2T4 decreases significantly if compared
to the parent compound, thus underscoring the role played by spatial distribution of the four peripheral positive charges in regulating proteasome–ligand interactions. We think that our results may pave the way to further studies aimed at rationalizing the molecular basis of novel, and more sophisticated, proteasome regulatory mechanisms.
Introduction
Proteasomes are large multi-subunit proteolytic assemblies that play a key role in regulating intracellular protein homeostasis.1
The 26S proteasome consists of an empty cylindrical 20S proteolytic core particle (CP) capped by one or two 19S Regulatory
Particles (RP).2,3The CP is made of four packed rings with seven
members each: twoa subunit rings (antechambers) bordering two centralb subunit rings (catalytic chamber).4–6In the catalytic chamber there are two duplicates of three different proteolyti-cally activeb subunits, i.e. b1, b2 and b5, exhibiting caspase-like (PGPH-L), trypsin-like (T-L) and chymotrypsin-like (ChT-L) activity, respectively.7–9 The externala rings, despite displaying
a nearly at surface, present shallow grooves between the subunits which are involved in the interactions with the RP.10
Several structural evidences indicate that the N-terminal tails of thea subunits form a “gate” at the center of the a ring which prevents substrate access in the absence of the RP molecule.11,12
Thus, the assembly of the mature 26S proteasome is a tightly regulated and reversible process, which leads to conformational changes displacinga subunit tails and opening the gate.13The
mechanism is supposed to be shared by all the RPs to date identied (i.e., 19S, PA28, and PA200);14 however, the lack of
detailed structural analyses on the 19S and the 26S do not allow to draw unambiguous conclusions on the mechanistic features of the process. Conversely, it has been recently demonstrated that an equilibrium between active open and closed gate CP
aIstituto di Biostrutture e Bioimmagini– CNR UOS di Catania, Via P. Gaifami 18,
95126 Catania, Italy. E-mail: [email protected]
bDipartimento di Scienze Chimiche, Universit`a di Catania, Viale Andrea Doria 6,
95125 Catania, Italy. E-mail: [email protected]
cDipartimento di Scienze Cliniche e Medicina Traslazionale, Universit`a di Roma Tor
Vergata, Via Montpellier 1, I-00133 Roma, Italy. E-mail: [email protected]
dIstituto di Biostrutture e Bioimmagini, CNR, Via Mezzocannone 16, 80134 Napoli,
Italy
e
Dipartimento di Scienze e Tecnologie Ambientali, Biologiche e Farmaceutiche, Seconda Universit`a degli Studi Napoli, Via Vivaldi 43, 81100, Caserta, Italy. E-mail: [email protected]
fDipartimento di Farmacia Universit`a di Napoli“Federico II”, Via D. Montesano, 49
I-80131 Napoli, Italy. E-mail: [email protected]
† Electronic supplementary information (ESI) available: It is reporting computational details, semi-log plots of residual CP activities, Molprobity results. See DOI: 10.1039/c5sc03312h
Cite this:Chem. Sci., 2016, 7, 1286
Received 3rd September 2015 Accepted 6th November 2015 DOI: 10.1039/c5sc03312h www.rsc.org/chemicalscience
Chemical
Science
EDGE ARTICLE
conformation exists in vitro even in absence of the RP;15hence,
several mechanisms are supposed to contribute to 20S protea-some gate regulation. The modulation of this phenomenon is a crucial point, since it is widely recognized that the RP-free 20S proteasome, which is supposed to exert a precise biological role by degrading oxidized and misfolded substrates regardless of the poly-ubiquitin tag, is present in a signicant amount inside the cell.16Therefore, in view of the relevant increase of the
intracel-lular 20S proteasome upon oxidative stress, its inhibition may bring about the accumulation of misfolded proteins and reactive oxygen species,17–19 and, ultimately, of proapoptotic factors
leading to cell-cycle arrest and cell death.20–22These evidences have prompted intensive research efforts over the last decade, which converged in the FDA approval of bortezomib, a dipeptide boronate competitive proteasome inhibitor, for the treatment of several hematological tumors and in particular multiple myeloma.23,24Unfortunately, the use of bortezomib as an
anti-cancer agent has severe limitations due to its off-target activity including remarkable side effects and drug resistance phenomena.25,26 Although the molecular mechanisms
accounting for the acquired resistance to bortezomib are not fully elucidated, several adaptive mutations harboring the binding site of bortezomib have been identied.27Therefore, the
search of new molecules, that allosterically affect proteasome activity by binding to sites far away from the catalytic centers, may be regarded as a promising antitumor strategy.28,29 In
a recent report some of us demonstrated that the cationic porphyrin meso-tetrakis(4-N-methylpyridyl)-porphyrin (H2T4)
inhibits, reversibly and with similar potencies, the three pro-teasome activities.30 Later on, biochemical studies have
conrmed our ndings by providing direct evidence that heme may bind to and inhibit the proteasome in a mouse embryo broblast cell line.31Beside their antiproteasome activity, there
are several reasons that make porphyrins attractive drug candi-dates for cancer treatment. As an example, porphyrins are extensively employed as photosensitizers in photodynamic therapy (PDT).32Moreover, porphyrins are extensively studied for
their capacity to bind G-quadruplex rich telomeric DNA struc-tures.33However, the lack of information about the relationship
between the molecular structure of cationic porphyrins and their antiproteasome activity is a major drawback to develop novel and more active porphyrin derivatives. Here, enzymatic assays, UV stoppedow, NMR experiments and computational approaches have been employed to get details about the binding and inhi-bition mechanism of the title compound – H2T4 – and two
isomers, ortho-H2T4 and meta-H2T4. The ability of H2T4 and its
two variants to interact with thea subunits of the CP correlates with the observed binding kinetics and inhibition potencies of the investigated molecules. Cationic porphyrins, may thus represent a new class of inhibitors targeting the gate of the CP.
Results and discussion
In vitro CP inhibition assay of H2T4 and of itsortho- and
meta- isomers
In a previous work,30we have found that H
2T4 inhibits rabbit
proteasome activity both in cell lysates and in puried CP
preparations. In the present study we have used a commercially puried human 20S proteasome, in order to avoid eventual interactions of porphyrins with possible contaminants of a cell extract. Indeed, the results obtained with isolated 20S protea-some, reported in Fig. 1, indicate that, independently from the proteasome source, H2T4 inhibits all the three proteolytic CP
activities with similar potency.
Moreover, it is worth recalling that previously reported data evidenced that the gradual decrease of the number of proton-ated pyridine rings from four to two (Fig. 2) corresponded to a parallel drastic reduction of the inhibitory potency (H2T4 >
tris-T4 > cis-T4).30
To better understand the effect of the spatial distribution of cationic charges on the activity of porphyrins we compared the IC50values of H2T4 with those of the ortho-H2T4 and meta-H2T4
isomeric forms. Fig. 3 reports the comparison of the IC50values
of H2T4 and its ortho- and meta- derivatives calculated for the
ChT-L, T-L, and PGPH-L CP activity. Dose–response plots were tted with eqn (2) (see panel B of Fig. S1 and Table S1 in ESI†). All three H2T4 variants retain a signicant potency in inhibiting
the PGPH-L activity, whereas for the ChT-L and T-L activities the inhibitory potency of ortho- and meta-isomers decreases signicantly. This is particularly evident for the ortho-H2T4
isomer, which inhibits the ChT-L and T-L activities much less efficiently (IC50¼ 2.59 mM and 2.47 mM, respectively) than the
other two isomeric forms (Fig. 1 in the ESI†). Altogether our previous and present results on the proteasome-inhibiting activity of cationic porphyrins underline the key role played by the positive charged nitrogen atoms on the pyridine rings. Indeed, both the number and the position of the positive charges change the inhibitory potency. In particular, the varia-tion of the spatial orientavaria-tion of the four charges, by varying the position of the nitrogen atom on the pyridine ring, differently affects the three proteasome catalytic activities. In this regard, the proteasome-inhibiting activity of the ortho- and meta-analogues of H2T4 may help to discern some simple rules for
structure–activity relationship (SAR) analysis: (1) the position of positively charged substituents is not relevant for the inhibition of the PGPH-L activity of the 20S proteasome; (2) by contrast,
Fig. 1 Normalized concentration–response plot for H2T4-mediated
inhibition of ChT-L, T-L and PGPH-L residual activities of 20S pro-teasome. The same data as a semilog plot are reported in panel A of Fig. S1 in (ESI†).
ChT-L and T-L activity are inhibited more effectively as the positively charged substituents are moved far away from the central ring with the highest activity when they arexed in the para position of the pyridine ring (Fig. 2).
Aer this initial screening of the antiproteasome efficiency of the three derivatives, we focused our attention on the inhi-bition mechanism of the parent molecule – H2T4-which
exhibited the lowest IC50values of all studied compounds.
NMR analysis of H2T4/proteasome complexes
In order to identify the functional groups involved in the H2
T4-CP interaction we have set up a ligand-detected NMR screening technique, such as Saturation Transfer Difference (STD) exper-iments. Initially, the structure of the ligand dissolved in the buffer (20 mM Tris–HCl, pH 7.6, 150 mM NaCl) has been analyzed by 1H NMR spectroscopy. In particular, H2T4 1H
chemical shi assignment has been obtained as it follows: 9.09 (8H, s, broad, Hpyr) 9.27 (8H, d, N-methylpyridine-Hm), 8.89 (8H, d, Ho), 3.69 (12H, s, N-methylpyridine-CH3). Aerwards, 860 nM of 20S proteasome was added to 172
mM of H2T4 (protein : ligand ratio of 1 : 200). Fig. 4a and b and
Fig. 4a0and 4b0clearly show that the1H H2T4 chemical shis do
not signicantly changes upon the addition of 20S proteasome, except for the pyrrole ring resonances, which display shis downeld (from 9.09 to 9.13 ppm); on the other hand, H2T4
resonances experience an overall so line broadening. These two effects, line broadening and chemical shi perturbations, which occurred on the proton of the same residues, conrmed the occurrence of H2T4 binding to the proteasome. To
struc-turally characterize the H2T4–20S proteasome interaction, we
apply STD (Saturation Transfer Difference) NMR spectroscopy assay34–41(Fig. 4c and c0).
The saturation affects all porphyrin resonances, conrming the binding to the CP. To identify the ligand moiety more closely interacting with the protein, we evaluated and compared the saturation effects of the individual H2T4 proton resonances
(ISTD¼ I0 Isat).42,43The signal showing the largest ISTD/I0value,
the methyl group of the N-methyl pyridine, was normalized to 100% (Fig. 5); the relative degree of saturation of the individual protons, normalized to that of the methyl group of the N-methyl pyridine, can be then used to compare the STD effect. Overall,
Fig. 2 Structures of the cationic porphyrins derivatives matter of analysis in our previous30and present studies.
Fig. 3 Comparison of the IC50 values of H2T4 –
para-tetra(4-N-methyil-pyridyl) porphyrin-,meta-H2T4–
meso-tetra(3-N-methyil-pyridyl) porphyrin– and ortho-H2T4– eso-tetra(2-N-methyil-pyridyl)
porphyrin– determined for the ChT-L (black), T-L (red) and PGPH-L (blue) peptidase activities of the CP.
Fig. 4 Low-field (left) and high-field (right)1H NMR spectra of H 2T4
(172mM) (a) in Tris–HCl (20 mM) pH 7.6, NaCl (150 mM) solution, and (b) in the presence of 20S (860 nM); (c)1H NMR STD spectrum of H2T4
in the presence of 20S (860 nM).1H chemical shift assignment has been reported for H2T4 in presence of 20S proteasome.
Fig. 5 Diagram showing the relative STD intensities for H2T4. For the
epitope mapping analysis, the STD integrals of the individual protons of H2T4 are referenced to the strongest STD signal, which is assigned to
the STD intensities show that the methyl-pyridine ring plays the major role in the protein binding, mainly through its ammo-nium moiety. On the other hand, the STD data indicate that the pyrrole ring is likely less directly involved in the recognition of the 20S proteasome. Accordingly, upon proteasome addition, the N-methyl pyridine resonances appear broadened, thus envisaging an intermediate exchange, whereas the pyrrole signals are shied, being in a fast exchange regime. As a control, a STD experiment with no added CP was also carried out; under these conditions, no signals were detected in the STD spectrum, suggesting that the saturation in the presence of 20S proteasome does not originate from non-specic interactions. Kinetic characterization of the effect of H2T4 on the ChT-L
activity of the CP
Analysis of enzyme kinetics may provide valuable insights in the understanding of inhibition mechanisms. To this aim, we have addressed the inhibition mechanism of the title compound H2T4 by determining the ChT-L activity of the CP at different
concentrations of substrate and inhibitor. Since H2T4 inhibits
all the three proteolytic functions of the CP with a similar potency, one can assume that the inhibition mechanism is similar for all enzymatic activities.
The Lineweaver–Burk plot shown in Fig. 6 indicates that H2T4 inhibits the CP by a competitive mechanism. Therefore,
the inhibitor affinity constant KIcan be obtained bytting data
according to eqn (4a) (see Experimental); the obtained value is KI¼ 3.6 (0.5) 107M.
The kinetic parameters of H2T4 binding to the CP have also
been determined as a function of time (as described in Exper-imental). Fig. 7 reports the time evolution of theuorescent product obtained by mixing at t¼ 0 both the substrate (100 mM nal concentration) and H2T4 at different concentrations, as
reported in the gure legend; the uorescent signal was
transformed into molar concentrations ofuorescent product through a calibration curve resulting from the complete pro-cessing of known concentration of substrate.
The continuous lines correspond to the simulation curve at different indicated concentrations of H2T4, employing Km¼ 7.2
105M, k
cat¼ 10 s1(i.e., the values of Kmand kcatobtained
from steady-state measurements, see above), and applying values of k+Iand of kI, which were constrained to the obtained
KI ¼ (kI/k+I). The resulting kinetic parameters for the H2T4
binding are k+I¼ 1.3 104M1s1and kI¼ 4.7 103s1,
indicating that H2T4 porphyrin behaves as a fast-reacting
inhibitor of the 20S proteasome, since the inhibition appears evident already few seconds aer the inhibitor addition. Molecular modeling
In collaboration with Prof. M. Groll from the Technische Uni-versit¨at M¨unchen, Germany, X-ray analysis have been carried out by soaking yeast 20S proteasome crystals with H2T4.
Notably, the porphyrin turned out to have a high propensity to self-aggregate in the presence of the crystallization buffer. Thus, the ligand concentration in the crystallization drop was inni-tesimal. Aer a soaking time of 24 hours, a dataset was recorded to 2.8 ˚A resolution, but the FO-FC-electron density map did not display any striking features that could be interpreted as a porphyrin core (personal communication). Therefore, in order to investigate the molecular details of the interaction of H2T4
with the human 20S proteasome, we performed molecular modeling studies. Firstly, the available experimentally deter-mined structures of 20S were downloaded from the Brookhaven Protein Data Bank (http://www.rcsb.org) and subjected to a structural and bioinformatics analysis. Then, an homology model of the human 20S proteasomea subunits was built (for details see Experimental in the ESI†). It must be mentioned here that, aer the completion of our molecular modeling studies, the X-ray structure of the human 20S proteasome was solved44
Fig. 6 Double reciprocal Lineweaver–Burk plot of substrate enzy-matic 20S processing in the absence of inhibitor (B) and in the presence of 0.5mM (), 1 mM (+), 3 mM (*), 10 mM (5) and 20 mM inhibitor (4) for various concentrations of the substrate Suc-LLVY-AMC (from 6 to 100mM).
Fig. 7 Kinetics offluorescent product formation, obtained by mixing 2 nM 20S proteasome, 100mM of substrate Suc-LLVY-AMC and different concentrations of H2T4 porphyrin, namely 0 (B), 0.3 mM () 3 mM (+)
and deposited in the Brookhaven Protein Data Bank (PDB IDs: 4R3O). Therefore, we were able to check the validity of our molecular model against the X-ray structure. Structural parameters were compared using structure evaluator soware40
and a summary of the results is reported in Table S2 in ESI.† We calculated the degree of similarity between the model and the X-ray structure and it resulted a 100% match of secondary struc-tures (i.e., helices, turns,b-strands) and an RMSD value on all alpha carbons of 0.61 ˚A (Fig. S2 in ESI†). Obtained results support the reliability of our molecular model and its use in the subsequent dynamic docking studies. Our previous30 and
present data underlined the crucial role played on inhibitory activity by the number and relative positioning of the proton-ated pyridine nitrogen atoms. Accordingly, we generproton-ated a 3-D pharmacophore model of the most active compound H2T4 by
assuming the four protonated nitrogen atoms as key interaction points and calculating their inter-atomic distances in the H2T4
X-ray structures present in the Cambridge Crystallographic Databank (CSD) (Fig. 8A; Table S3 in ESI†). Then, we evaluated the ability of the human 20S proteasome to accommodate the rigid, planar and positively charged pharmacophore of H2T4 by
mapping the spatial positioning of negatively charged amino
acids on the protein surface, as well as in the known functional and inhibitor binding sites (Table S4 in ESI†). The analysis was performed considering all the experimentally determined conformational states (closed, open and semi-closed) and the sequence homologies among the different species were calcu-lated using the PROMALS3D45 server (http://
www.prodata.swmed.edu/promals3d/promals3d.ph). Obtained results excluded all the 20S catalytic sites and prompted us to hypothesize the substrate gate (closed state) as the most prob-able H2T4 binding site, for the competitive inhibition of 20S
catalytic activities (Fig. 8B).
This region is lined by specic residues belonging to the N-terminal tails of thea subunits. Going into details, with the exception of a2, all the N-terminal tails contain an aromatic residue (Phe ata1 and Tyr at a3–a7) followed by a negatively charged residue (Asp); altogether, they form a sort of ring which widens up during gate opening (Fig. S3 in ESI†). Both residues are conserved among the considered species (i.e., human, mouse, yeast, Thermoplasma acidophilum), being involved in gate opening and functioning.46–48Interestingly, we found that
the negatively charged Asp residues on the N-terminal tails of a1, a5, a6, and a7 display suitable spatial positioning (in the “closed state”) for a possible interaction with the positively charged H2T4 nitrogens (Fig. 8C). In addition, the four adjacent
aromatic residues could assist the binding by establishing polarized p–p interaction with the pyridine rings (Fig. 8D). Thus, H2T4 was positioned at its putative binding site on the
human proteasome structural model and the obtained complex was used as starting structure for the subsequent fully dynamic docking calculations. Although our docking protocol formally requires a starting complex, it has to be underlined that all protein atoms included in the binding domain area are le free to move (conformational search, rotation, and translation). To fully explore all possible binding sites/modes, the binding domain area was dened as the whole a ring of the human 20S proteasome. Firstly, a Monte Carlo/minimization approach for the random generation of a maximum of 20 acceptable complexes was used. To ensure a wide variance of the input structures to be successively minimized, an energy tolerance value of 106kcal mol1from the previous structure was used. Aer the energy minimization step, the energy test, with an energy range of 50 kcal mol1, and a structure similarity check (rms tolerance ¼ 0.3 kcal ˚A1) was applied to select the 20 acceptable structures. The resulting complexes are then subject to simulated annealing (SA) calculations. In SA the temperature is altered in time increments from an initial (500 K) to anal (300 K) temperature. The temperature of 500 K was applied with the aim of surmounting torsional barriers, thus allowing a full rearrangement of the ligand and the protein (see Experimental section for details). During all calculations, constraints were applied only to hydrogen bonds (a-helices) and f and 4 torsion angles (b-sheets) of some structured regions, by using different force constant values according to the results of secondary structure prediction calculations (http:// www.predictprotein.org/). In particular, a force constant of 100, 10, and 1 kcal mol1 was applied to hydrogen bonds (a-helices) and f and 4 torsion angles (b-sheets) with high,
Fig. 8 (A) H2T4 pharmacophore and related inter-atomic distances,
the experimentally determined structure of H2T4 (CSD code: OBOZAI)
is displayed in stick with the pyridine nitrogen atoms evidenced in CPK. (B) Top view of the human 20S proteasome (PDB ID: 4R3O); only the a ring is shown for clarity of presentation. (C) Zoom of the top view of the human 20S proteasome (PDB ID: 4R3O); the four Asp residues are displayed as CPK and colored in red and the suitable inter-residue distances for a possible interaction with the H2T4 pharmacophore are
reported. (D) Positioning of H2T4 in the putative binding site; the
potential interactions between H2T4 and the 20S proteasome are
shown: the amino acid residues involved in ionic and cation-p inter-actions are displayed as CPK and colored in red and yellow, respec-tively. H2T4 is colored by atom type (C: green; N: blue). Thea subunits
are colored in pink (a1), orange (a2), brown (a3), light green (a4), cyan (a5), magenta (a6), and gray (a7).
medium, and low secondary structure prediction score, respectively. The rest of the atoms of the protein was le free of constraints on its movement. The quality of the resulting complexes was then assessed by using structure evaluator soware40 (see Table S5 in ESI†). Docking studies disclosed
some interesting results about H2T4 ability to inhibit the 20S
proteasome acting as a“plug” of the CP entrance gate. Firstly, all the complexes generated by our dynamic docking protocol, either by Monte Carlo and by SA calculations, always presented H2T4 bound to the entrance gate of the channel, interacting
with the identied N-terminal residues of the a subunits (Table S6 in ESI† and Fig. 9).
At the end of the docking protocol, it turned out that the lowest energy complex is also the one characterized by the most favourable non-bond interaction energy, so it was selected as the best docked complex (Fig. 10, Table S6 in ESI†).
Interestingly, the H2T4 binding mode calculated by the
Monte Carlo docking procedure signicantly changed aer it is subjected to SA calculations (Fig. 10A vs. B). Indeed, in the Monte Carlo complex H2T4 binds to the gate assuming a
posi-tion perpendicular to thea ring plane, establishing interactions with the negatively charged residue D9 (a5) and D7 (a6), and the aromatic residues F9 (a1), Y8 (a5) and Y6 (a6) (Fig. 10A). When this complex is subjected to the SA procedure, a much more stable ligand–protein complex is achieved, characterized by the most favourable non bond interaction energy (complex 2 in Table S6 in ESI†). In this last complex, H2T4 binds parallel to
thea subunits plane and interacts with the whole cluster of the identied functional residues at the N-terminal tails of a1, a3, a5, a6, and a7 (Fig. 10B and Table S7 in ESI†). In order to validate our 20S–H2T4 interaction model, meta-H2T4 and
ortho-H2T4 were subjected to the same dynamic docking protocol
applied to H2T4. Obtained results evidenced that, during the
docking simulation, contrarily to what observed for H2T4
(Fig. 9) and in agreement with the varied spatial positioning of their protonated nitrogen atoms, meta-H2T4 and ortho-H2T4 do
not remain bound to the identied N-terminal Asp residues at the 20S gate (Fig. S4 in ESI†). Moreover, in the complex char-acterized by the lowest non bond interaction energy (Fig. S5 in ESI†), they bind at a side of the gate, interacting only with F9 (a1) and Y8 (a5), meta-H2T4, and Y6 (a6), ortho-H2T4.
Accord-ingly, meta-H2T4 and ortho-H2T4 showed an overall lower
inhibitory potency with respect to H2T4 and inhibited the three
catalytic activities to different extents (Fig. 3). Interestingly, ortho-H2T4 interacts with negatively charged residues, in the
groove between a5 and a6 subunits, reported to be directly involved in the binding to the positively charged tails of the 20S regulatory proteins (RPs).49–52Thus, the difference in the puta-tive binding modes of the three isomers seems to reect their different inhibitory properties. Indeed, the modulation of the catalytic activity by meta-H2T4 and ortho-H2T4 may occur
through a more sophisticated interference with proteasome
Fig. 9 Top view of the dynamic docking results obtained for H2T4: (A)
Monte Carlo; (B) SA. The backbone of the starting complex is displayed as solid ribbons and colored in gray, the one of the calculated complexes is displayed as line ribbons and colored in orange. The cluster of negatively charged residues at the entrance gate of the CP channel is displayed as CPK and colored in gray (starting complexes) and red (calculated complexes). The porphyrin ligands are colored by atom type (C: green; N: blue; H: white) and displayed as CPK. In (A) the a subunits and the position of the catalytic b subunits are labelled.
Fig. 10 Results obtained by the dynamic docking procedure. H2
T4-20S Monte Carlo (A) and SA (B) complexes. Thea subunits are dis-played as line ribbons and colored in gray. H2T4 is displayed as CPK
and colored by atom type (C: green; N: blue). The amino acid residues involved in ionic and cation-p interactions are displayed as CPK and colored in red and yellow, respectively. Protein van der Waals volume is displayed as transparent surface (bottom representation).
allosteric regulation with respect to H2T4, being the three
pro-teasome catalytic activities connected to the opening of the gate through different allosteric mechanisms. At this regard, it is worth of note that the PGPH-L activity is the most sensitive to the presence of activators such as PA28 or SDS53which open the
gate of the 20S proteasome. This is ascribable to the more pronounced effects that gating mechanisms may have on the PGPH-L activity11and therefore, fully reconcile with the
inhi-bition model proposed in the current work.
Overall, molecular modeling results allow us to hypothesize that H2T4 binds the predominant54closed/latent conformation
of 20S proteasome at the substrate gate, and that the initial binding is then followed by a conformational change of the inhibitor–enzyme complex (i.e., induced t mechanism)55
resulting in the formation of a more stable complex. On the basis of these results, it is plausible to postulate that H2T4
binding to the 20Sa subunits could affect the catalytic activities not only by clogging the gate and, consequently, the entry of all substrates, but also by affecting the conformational equilibrium of the 20S human proteasome and inducing a shi in the equilibrium of the “open-to-close” structural transition. Following this hypothesis and considering the results of the subsequent stopped-ow kinetic studies (see next paragraph), we can propose that the binding of H2T4 induces a
conforma-tional change that may be related to the availability of another binding site for H2T4 on the 20S proteasome. In this view, it is
worth of note that our structural and bioinformatics analysis identied a cluster of negatively charged residues showing suitable inter-atomic distances for a possible interaction with the porphyrin pharmacophore at the interface betweena1-b1, a2-b2, and a5-b5 subunits (Fig. S6 and Table S4 in ESI†). These residues are involved in the modulation of enzyme conforma-tions48 and were already reported as the binding site of the
noncompetitive inhibitor chloroquine.56 Dynamic docking
calculations, performed considering this second binding site (for details see Experimental section in ESI†), put in evidence the ability of H2T4 to interact with the identied negatively
charged residues (Fig. S7 in ESI†), but with a lower affinity with respect to that showed for the gate, as proved by the calculated non-bond interaction energies (Table S8 in ESI†). Indeed, as reported above, the 20S gate is characterized not only by a cluster of negatively charged residues but also by a cluster of aromatic residues, that contribute to the interaction with H2T4
(Table S7 in ESI†). In any case, it has to be underlined that, according to its non-competitive nature, the occupation of this site cannot be directly involved in the observed competitive enzyme inhibition by H2T4.
Kinetic investigation of the interaction between H2T4 and 20S
proteasome by stopped-ow UV spectroscopy
In order to nd a closer correlation between experimental observations and molecular modeling we have carried out a kinetic investigation of the interaction between H2T4 and 20S
proteasome by stopped-ow. Fig. 11 shows the absorption changes at 421 nm, which turned out to be the wavelength at which the process displayed the largest spectral changes;
however, the behavior observed was essentially identical also at other wavelengths, though showing some variations for the amplitudes of different processes. Data have been analyzed according to the following equation
DODobs¼
Xi¼r i¼1
DODiexpðkitÞ (1)
whereDODobsis the observed optical density change at a given
time, r is the total number of exponentials during the reaction time course,DODiis the optical density change corresponding
to the exponential i, ki is the rate constant associated to the
exponential i, t is the time.
The reaction between the 20S proteasome and the H2T4
porphyrin displayed a decrease of absorption in the Soret region, as expected from previous observations at equilibrium. However, it is important to outline that kinetics reported in Fig. 11 are characterized by a multiphasic pattern with different patterns of concentration dependence of the rate constants. Fitting kinetic curves according to eqn (1) indicated that the process can be accounted for by (at least) three exponentials, two of which (namely the faster ones) turned out to be bimo-lecular; their values are reported in Table 1. It appears evident as the process is characterized by a faster interaction step, fol-lowed by a somewhat slower event (which displays the same
Fig. 11 Optical density changes at 421 nm for the mixing at 37C of 1 nM 20S proteasome with different concentrations of H2T4, namely 1
mM (curve a), 2 mM (curve b) and 5 mM (curve c). Continuous lines correspond to the non-linear least-squaresfitting of data with two exponentials according to eqn (1), employing parameters reported in Table 1.
Table 1 Kinetic parameters for the reaction of H2T4 with the 20S
proteasome onk 1(M1s1) 5.65 (0.74) 104 offk 1(s1) 0 onk 2(M1s1) 1.28 (0.31) 104 offk 2(s1) 0 k3(s1) 3.23 (0.81) 103
bimolecular rate constant observed for the competitive inhibi-tory process, see Fig. 6). Therefore, on the basis of molecular modeling studies, these kinetic data can be accounted for by assuming that there is a rst faster interaction with the N-terminal tails of thea subunits, characterized by k1, likely
cor-responding to therst encountering of the porphyrin with the a-subunits plane in a perpendicular geometry (Fig. 10A). This event is then followed by a tighter binding to the closed CP gate, characterized by k2, which can be referred to the formation of
the more stable porphyrin–protein complex characterized by a parallel geometry (Fig. 10B). Notably this k2value reconciles
with that obtained from the activity assays (k+I¼ 1.3 104M1
s1). The valuesz0 for the dissociation rate constants (Table 1) simply means that the rate is very slow and the interaction is very strong.
The absorption change is also characterized by an additional process with k3¼ 3.23 (0.81) 103s1, whose rate is
inde-pendent on porphyrin concentration, likely corresponding to a subsequent slower binding of porphyrins to the 20S protea-some at site(s) topologically distinct from that of therst two events. The observed kinetic process appears to be rate-limited by a conformational change, which follows and depends on the porphyrin binding to the CP gate. This slower process, as sug-gested by molecular modeling studies, might be related to the availability of an additional binding site for H2T4 on the 20S
proteasome (see above). As a whole, kinetic results and computational simulations appear to support each other, strengthening the proposed mechanism of interaction between H2T4 and the 20S proteasome, reinforcing the idea that this
porphyrin not only acts as a “plug” of the CP gate,57 but it
behaves as a quaternary effector through an additional slow induced-t mechanism, which brings about a conformational shi of the 20S proteasome, stabilizing the “closed” structure.
Experimental
Chemicals
Puried human 20S proteasome and the uorogenic substrates Suc-LLVY-AMC, Z-Leu-Leu-Glu-AMC, and Ac-Arg-Leu-Arg-AMC used to test the ChT-L, PGPH-L, and T-L activity, respectively, were purchased from Boston Biochem (Cambridge, MA, USA). Meso-tetrakis(4-N-methylpyridyl) porphyrin (para-H2T4 or
H2T4), meso-tetra(2-N-metil-pyridyl) porphyrin (ortho-H2T4),
and meso-tetra(3-N-metil-pyridyl) porphyrin (meta-H2T4) were
purchased from Midcentury.
Proteasome assays
Proteasome activity assays were performed in vitro by mixing 20S proteasome (2 nM) with 100mM of the proper uorogenic peptide in the assay buffer (i.e., 25 mM HEPES, 0.5 mM EDTA, pH 7.6) in a 384 multiwell black plate. The released AMC uo-rescence was recorded at 440 nm (excitation at 360 nm) for 20 min, that is a time interval over which linearity was observed in auorescence plate reader (Varioskan, Thermo). A minimum of three replicates were performed for each data point. Data are expressed as normalized percentages of residual activity
considering the slope of the control (uorogenic peptide/pro-teasome in the absence of inhibitors) as 100% of propeptide/pro-teasome activity. Dose–response plots of the residual proteasome activity in the presence of increasing concentration of inhibitor provides a quantitative estimate of its potency. The IC50 is
dened as the concentration of the inhibitor which causes 50% reduction of activity and it is thus calculated from the x-axis value of the dose–response plot occurring at a fractional activity of 50%. The estimation of the IC50is based on a nonlineart
with the equation:
vð%Þ ¼ 100 IC50
½I þ IC50
¼ 100
1þ 10ðlog½Ilog IC50Þ (2)
The midpoint of this function occurs at a fractional velocity value of 50%, corresponding to half inhibition of the target enzyme. The IC50values and their standard errors were deduced
from the tting. Data relative to CP activities at different substrate/inhibitor concentrations have been analyzed by a double reciprocal Lineweaver–Burk plot, according to the following equation: ½E0 v ¼ Km kcat 1 ½Sþ 1 kcat (3) where [E0] is the enzyme concentration, n is the observed
velocity (expressed as moles of substrate cleaved per time interval unit) and [S] is the substrate concentration. The resulting catalytic parameter are Km, corresponding to the
apparent affinity constant (or Michaelis–Menten constant) of substrate for the free enzyme (to form the ES complex), and kcat,
corresponding to the velocity of the rate-limiting step during the enzymatic activity. Inhibition assays of 20S proteasome ChT-L activity have been carried out by incubating the 20S proteasome with increasing concentrations (0.1–3 mM) of inhibitor for 30 min at 37C; ChT-L activity was then tested in the same way as described above by adding increasing concentrations of the substrate. The mechanism of inhibition of the 20S proteasome ChT-L activity by the inhibitors was analyzed by modied double-reciprocal plot, as from the following equations for the different inhibitory mechanisms:
½E0 v ¼ Km 1þ½I KI kcat 1 ½Sþ 1 kcat (4a)
for a competitive inhibition ½E0 v ¼ Km 1þ½I KI kcat ½S1 þ 1þ ½I aKI kcat (4b)
for non-competitive inhibition, and ½E0 v ¼ Km kcat ½S1 þ 1þ½I KI kcat (4c)
for un-competitive inhibition, where KIis the inhibitor affinity
constant anda is the linkage factor, representing the reciprocal effect on the substrate and inhibitor affinity constants in the non-competitive inhibition and being related to the simulta-neous presence in their respective binding sites. Whena is very large, binding of inhibitor severely impairs binding of the substrate and the non-competitive inhibition becomes closely similar to competitive inhibition (since in the second factor of eqn (4b) (1 + [I]/(a KI))z 1, as in eqn (4a)); conversely, when
a is very small the non-competitive model becomes nearly identical to an un-competitive model eqn (4c), since in eqn (4b) (1 + [I]/KI) (1 + [I]/(a KI)). All curvetting and statistical
analysis were carried out using the Non Linear Fitting Tool (NLFit) and MatLab (The Math works Inc., Natick, MA, USA). The parametric datatting was based on nonlinear regression and the method of least squares. Model discrimination and choice was based on the goodness oft. The goodness of t was evaluated by visual examination of the tted curves, 95% condence bounds for the tted coefficients and statistical analysis for determining the square of the multiple correlation coefficient (R2).
The interaction of H2T4 with 20S proteasome has been also
investigated from the pre-steady-state kinetic standpoint. Briey, these assays were performed without incubation time, by direct addition of different concentrations of H2T4 in
samples where the Ch-L activity was already being monitored. The extent and the rate of inhibition were determined by following the time-dependentuorescent signal of the cleaved substrate for 25 additional minutes. Data have been analyzed according to a second-order Runge–Kutta simulation procedure based on a simple competitive inhibition mechanism, such as in Scheme 1.
The time evolution of each species reported in Scheme 1 was calculated at discrete time intervals dt. Thus, assuming that equilibrium between [E] and [ES] is very fast (which is the main assumption of the application of the Michaelis–Menten equa-tion), the time evolution of the product formation is described by the following equation:
[P]t¼x¼ [P]t¼x1+ (kcatKm[S]t¼x1[E]t¼x1)dt (5a)
which is regulated by the following two equations:
[E]t¼x¼ [E]t¼x1+ (kI[EI]t¼x1 k+I[E]t¼x1[I]t¼x1)dt (5b)
[EI]t¼x¼ [EI]t¼x1+ (k+I[E]t¼x1[I]t¼x1 kI[EI]t¼x1)dt (5c)
with all populations reequilibrated at each time step. NMR measurements
Samples used for NMR experiments were prepared using a solution of 1.03mM of 20S in Tris–HCl (20 mM) pH 7.6, NaCl (150 mM) and DTT (1 mM) and a solution containing the ligand at a concentration of 18.1 mM in DMSO-d6. The NMR tubes were prepared by adding 250mL of the 20S solution, 2.9 mL of H2T4 and 47.1mL of D2O to a 3 mm NMR tube. Thenal tube
contained 0.860mM of 20S and 172 mM of the ligand which corresponds to a protein/ligand ratio of 1 : 200. The use of 3 mm NMR tubes is particularly useful in this type of study since it allows to minimize the amount of protein used and to reduce salt effects induced by the buffer. NMR experiments were acquired at 298 K on a Varian Inova 600 MHz spectrometer, equipped with a cold probe optimized to detect 1H nucleus,
located at the Institute for the Biostructures and the Bioimages of the CNR in Naples. The1H NMR signal assignment was ob-tained from the analysis of the spectra assisted by the theoret-ical prediction based in the molecular structure using ChemAxon soware (http://www.chemaxon.com). Saturation Transfer Difference (STD) spectra34,35 were acquired using
a series of equally spaced 50 ms Gaussian-shaped pulses for selective saturation, with 1 ms delay between the pulses and a total saturation time of 2 s. The frequency of the protein (on-resonance) saturation was set to the protein1H NMR signals in the low frequency region1 ppm. The STD factor were calcu-lated as ASTD¼ (I0 Isat)/I0¼ ISTD/I0where I0is the intensity of
the signal in the reference experiment and Isatis the intensity of
the same signal in the saturated spectrum. The signal obtained with the strongest ISTD/I0 value was normalized to 100%. The
relative degree of saturation for the individual protons was used to compare the STD effect.36
Molecular modeling
Molecular modeling calculations were performed on SGI Origin 200 8XR12000 and E4 Server Twin 2 Dual Xeon—5520, equipped with two nodes. Each node: 2 Intel® Xeon® Quad-Core E5520—2.26 GHz, 36 GB RAM. The molecular modeling graphics were carried out on SGI Octane 2 workstations. Analysis of the steric and electronic properties of H2T4 and its
derivativesmeta-H2T4 andortho-H2T4
The experimentally determined structures of H2T4 (CSD codes:
IDEBUO, IDECAV, OBOZAI, PIGFIV, PUBCAR, PUBCEV, PUB-CIZ, SIKJOL and TEDMOF) were downloaded from the Cam-bridge Structural Database (CSD) using the CSDS (CamCam-bridge Structural Database System) soware Conquest 1.16. meta-H2T4
and ortho-H2T4 were built by modifying the experimentally
determined structure of H2T4 (CSD code: OBOZAI; Insight2005
Builder module). The apparent pKavalues of porphyrins were
calculated by using ACD/Percepta soware.37The compounds
were considered in their tetra-cationic form in all calculations performed as a consequence of the estimation of percentage of
Scheme 1 Schematic representation of the competitive enzyme inhibition mechanism.
neutral/ionized forms computed at pH 7.4 (physiological value) and pH 7.2 (cytoplasmic value) using the Handerson–Hassel-balch equation. Atomic potentials were assigned to the compounds using the CVFF force eld,38 while the partial
charges were assigned using the partial charges estimated by MNDO semi-empirical 1 SCF calculations.39 These structures
were analyzed using Insight 2005 (Accelrys, San Diego, CA).
Porphyrins/CP docking studies
The molecular models ofa1–a7 subunits of 20S human pro-teasome were built (for details see Experimental in the ESI†) and the obtained homology model was used for the docking studies. In particular, a docking methodology, which considers all the systems exible (i.e., ligand and protein), was used (Affinity, SA_Docking; (Insight2005, Accelrys, San Diego). Although in the subsequent dynamic docking protocol all the systems were perturbed by means of Monte Carlo and simulated annealing procedures, nevertheless, the dynamic docking procedure formally requires a reasonable starting structure. Accordingly, the putative starting complexes (H2T4/20S human
proteasomea1–a7 subunits; meta-H2T4/20S human proteasome
a1–a7 subunits and ortho-H2T4/20S human proteasomea1–a7
subunits) were subjected to a preliminary energy minimization (Steepest Descent algorithm, maximum RMS derivative¼ 1 kcal ˚A1;3 ¼ 1). Aer the docking procedure (for details see
Exper-imental section in ESI†), the complex with the best non-bond interaction energy was selected as structure representing the most probable binding mode. The selected complexes were checked for quality using Molprobity structure evaluator soware.40
Conclusions
It is known that the 20S proteasome can cleave a wide range of proteins and peptides without the assistance of ancillary RPs58
and the key role played by this ubiquitin-independent degra-dation pathway in maintaining cell homeostasis is increasingly recognized.59Furthermore, a tight control of protein
degrada-tion by the gating ends of the CP is vital for cell viability.60
Therefore, the search for molecules that can affect the dynamic mechanism of 20S CP gating is a highly promising strategy which may open unprecedented possibilities of regulating proteasome function. Our results suggest that cationic porphyrins are efficient and versatile CP gatekeepers. NMR studies show that the methyl-pyridyne moieties of H2T4 have
tighter contacts with the protein surface than the pyrrole rings and further support the hypothesis that electrostatic charges provide the major driving force in porphyrin/CP interaction. Accordingly, slight changes in the position of the positive charges, such as in the ortho- or meta-H2T4 analogues, imply
signicant effects not only on the potency of the molecules but also on their binding mode, selectively affecting the catalytic activities of human 20S proteasome. On these bases, future design strategies will include the rational modication of the number, nature, and spatial positioning of the protonated nitrogens attached to the porphyrin system. In addition, we
show that H2T4 may bind at least two sites of the CP, both
characterized by the presence of a cluster of negatively charged residues, playing a key role in modulating enzyme conforma-tions/functions. In therst, fast binding mode the porphyrin glides on and clogs up (with two temporally distinct dynamic processes) the CP gate, thus leading to the simultaneous inhi-bition of the ChT-L, T-L and PGPH-L activities. A second much slower event relates to the subsequent adhesion of the porphyrin to the CP external surface and, in particular, to the grooves which separate thea from the b subunits. This slower binding mode, which is not directly related to the stereochem-istry of the observed CP competitive inhibition by porphyrins, is nonetheless relevant, since it brings about a porphyrin-depen-dent conformational change of the 20S proteasome. As a whole, all data here presented lead to envisage a mechanism where H2T4“plugs” the CP gate, behaving as a competitive inhibitor,
even though, it is not a competitive inhibitor, since it does not bind to the catalytic site. This stems from the complex nature of the substrate interaction with proteasome, which is at the same time allosteric (a positive feedback loop connects the site of amide bond hydrolysis with the site of ingress/egress of substrates and products) and classical enzymatic (displaying a Michaelis–Menten behaviour at the catalytic site). Therefore, though displaying a competitive-like behaviour, porphyrin also interferes with the physiological quaternary equilibrium of the 20S proteasome between“open” and “closed” conformations, shiing it toward a “closed” structure with a consequent further enhancement of the energetic barrier for the gate opening of CP. These peculiar anti-proteasome properties, coupled with their versatile chemistry, propose cationic porphyrins as a novel class of CP conformational regulators with great potentiality as “lead” pharmacophores.
Acknowledgements
We gratefully acknowledge Prof. M. Groll for X-rays analysis of porphyrin/proteasome assemblies. This work was nancially supported by the Italian Ministry of University and Research (MIUR): FIRB 2012 RBFR12WB3W.
Notes and references
1 D. Finley, Annu. Rev. Biochem., 2009, 78, 477.
2 S. Murata, H. Yashiroda and K. Tanaka, Nat. Rev. Mol. Cell Biol., 2009, 10, 104.
3 N. Gallastegui and M. Groll, Trends Biochem. Sci., 2010, 35, 634.
4 O. Coux, K. Tanaka and A. L. Goldberg, Annu. Rev. Biochem., 1996, 65, 801.
5 W. Baumeister, J. Walz, F. Z¨uhl and E. Seem¨uller, Cell, 1998, 92, 367.
6 M. H. Glickman, D. M. Rubin, O. Coux, I. Wefes, G. Pfeifer, Z. Cjeka, W. Baumeister, V. A. Fried and D. Finley, Cell, 1998, 94(5), 6156.
7 C. S. Arendt and M. Hochstrasser, Proc. Natl. Acad. Sci. U. S. A., 1997, 94, 7156.
8 W. Heinemeyer, M. Fischer, T. Krimmer, U. Stachon and D. H. Wolf, J. Biol. Chem., 1997, 272, 25200.
9 A. F. Kisselev, A. Callard and A. L. Goldberg, J. Biol. Chem., 2006, 281, 8582.
10 M. Groll, L. Ditzel, J. L¨owe, D. Stock, M. Bochtler, H. D. Bartunik and R. Huber, Nature, 1997, 386, 463. 11 M. Groll, M. Bajorek, A. Kohler, L. Moroder, D. M. Rubin,
R. Huber, M. Glickman and D. Finley, Nat. Struct. Biol., 2000, 7, 1062.
12 M. Groll and R. Huber, Int. J. Biochem. Cell Biol., 2003, 5, 606. 13 A. Kohler, M. Bajorek, M. Groll, L. Moroder, D. M. Rubin, R. Huber, M. Glickman and D. Finley, Biochimie, 2001, 83, 325. 14 D. J. McConkey and K. Zhu, Drug Resist. Updates, 2008, 11, 164. 15 A. Lehmann, K. Jechow and C. Enenkel, EMBO Rep., 2008,
9(12), 1237.
16 G. Ben-Nissan and M. Sharon, Biomolecules, 2014, 4, 862. 17 E. A. Obeng, L. M. Carlson, D. M. Gutman, W. J. Harrington
Jr, K. P. Lee and L. H. Boise, Blood, 2006, 107(12), 4907. 18 S. T. Nawrocki, J. S. Carew, K. Dunner Jr, L. H. Boise,
P. J. Chiao, P. Huang, J. L. Abbruzzese and D. J. McConkey, Cancer Res., 2005, 65(24), 11510.
19 G. Bianchi, L. Oliva, P. Cascio, N. Pengo, F. Fontana, F. Cerruti, A. Orsi, E. Pasqualetto, A. Mezghrani, V. Calbi, G. Palladini, N. Giuliani, K. C. Anderson, R. Sitia and S. Cenci, Blood, 2009, 13(13), 3040.
20 L. R. Dick and P. E. Fleming, Drug Discovery Today, 2010, 15(5–6), 243.
21 E. M. Huber and M. Groll, Angew. Chem., Int. Ed. Engl., 2012, 51(35), 8708.
22 M. Bazzaro, M. K. Lee, A. Zoso, W. L. H. Stirling, A. Santillan, I.-M. Shih and R. B. S. Roden, Cancer Res., 2006, 66(7), 3754. 23 O. A. O'Connor, J. Wright, C. Moskowitz, J. Muzzy, B. MacGregor-Cortelli, M. Stubbleeld, D. Straus, C. Portlock, P. Hamlin, E. Choi, O. Dumetrescu, D. Esseltine, E. Trehu, J. Adams, D. Schenkein and A. D. Zelenetz, J. Clin. Oncol., 2005, 23(4), 676.
24 J. Wolf, P. G. Richardson, M. Schuster, A. LeBlanc, I. B. Walters and D. S. Battleman, Clin. Adv. Hematol. Oncol., 2008, 6(10), 755–760.
25 D. J. Kuhn, R. Z. Orlowski and C. C. Bjorklund, Curr. Cancer Drug Targets, 2011, 11(3), 285.
26 S. Kumar and S. V. Rajkumar, Blood, 2008, 112(6), 2177. 27 E. M. Huber, W. Heinemeyer and M. Groll, Structure, 2015,
23, 407–417.
28 X. Li, T. E. Wood, R. Sprangers, G. Jansen, N. E. Franke, X. Mao, X. Wang, Y. Zhang, S. E. Verbrugge, H. Adomat, Z. H. Li, S. Trudel, C. Chen, T. L. Religa, N. Jamal, H. Messner, J. Cloos, D. R. Rose, A. Navon, E. Guns, R. A. Batey, L. E. Kay and A. D. Schimmer, J. Natl. Cancer Inst., 2010, 102(14), 1069.
29 J. Li, M. Post, R. Volk, Y. Gao, M. Li, C. Metais, K. Sato, J. Tsai, W. Aird, R. D. Rosenberg, T. G. Hampton, F. Sellke, P. Carmeliet and M. Simons, Nat. Med., 2000, 6(1), 49. 30 A. M. Santoro, M. C. Lo Giudice, A. D'Urso, R. Lauceri,
R. Purrello and D. Milardi, J. Am. Chem. Soc., 2012, 134, 10451.
31 F. Vallelian, J. W. Deuel, L. Opitz, C. A. Schaer, M. Puglia, M. L¨onn, W. Engelsberger, S. Schauer, E. Karnaukhova, D. R. Spahn, R. Stocker, P. W. Buehler and D. J. Schaer, Cell Death Differ., 2015, 22(4), 597.
32 T. J. Dougherty, C. J. Gomer, B. W. Henderson, G. Jori, D. Kessel, M. Korbelik, J. Moan and Q. Peng, J. Natl. Cancer Inst., 1998, 90(12), 889.
33 V. Rapozzi, S. Zorzet, M. Zacchigna, E. Della Pietra, S. Cogoi and L. E. Xodo, Mol. Cancer, 2014, 13(75), 1.
34 J. Klages, M. Coles and H. Kessler, NMR-Based Screening: A Powerful Tool in Fragment-Based Drug Discovery, in Exploiting Chemical Diversity for Drug Discovery, ed. P. A. Bartlett and M. Entzeroth, RSC Publishing, Cambridge, 2006, ch. 12, pp. 263–290 and Mol. BioSyst., 2006, vol. 2, pp. 318–332.
35 B. Meyer and T. Peters, Angew. Chem., Int. Ed. Engl., 2003, 42(8), 864.
36 J. Klein, R. Meinecke, M. Mayer and B. Meyer, J. Am. Chem. Soc., 1999, 121(22), 5336.
37 ACD/Percepta, version 14.0.0, Advanced Chemistry Development, Inc., Toronto, ON, Canada, 2013, http:// www.acdlabs.com.
38 P. Dauber-Osguthorpe, V. A. Roberts, D. J. Osguthorpe, J. Wolff, M. Genest and A. T. Hagler, Proteins, 1998, 4, 31. 39 M. J. S. Dewar and W. Thiel, J. Am. Chem. Soc., 1977, 99, 4899. 40 I. W. Davis, A. Leaver-Fay, V. B. Chen, J. N. Block, G. J. Kapral, X. Wang, L. W. Murray, W. B. Arendall, J. Snoeyink, J. S. Richardson and D. C. Richardson, Nucleic Acids Res., 2007, 35, W375.
41 M. Mayer and B. Meyer, J. Am. Chem. Soc., 2001, 123, 6108. 42 J. Pons, N. Evrad-Todeschi, G. Bertho, J. Gharbi Benarous,
V. Tanchou, R. Benarous and J. P. Girault, Biochemistry, 2008, 47, 14.
43 L. Cigliano, L. de Rosa, D. Diana, R. Di Stasi, M. S. Spagnuolo, B. Maresca, R. Fattorusso and L. D. D'Andrea, RSC Adv., 2014, 4, 51353.
44 W. Harshbarger, C. Miller, C. Diedrich and J. Sacchettini, Structure, 2015, 23(2), 418.
45 J. Pei, B. H. Kim and N. V. Grishin, Nucleic Acids Res., 2008, 36, 2295.
46 J. Rabl, D. M. Smith, Y. Yu, S. C. Chang, A. L. Goldberg and Y. Cheng, Mol. Cell, 2008, 30(3), 360.
47 A. F¨orster, E. I. Masters, F. G. Whitby, H. Robinson and C. P. Hill, Mol. Cell, 2005, 18(5), 589.
48 A. M. Ruschak and L. E. Kay, Proc. Natl. Acad. Sci. U. S. A., 2012, 109(50), E3454.
49 F. G. Whitby, E. I. Masters, L. Kramer, J. R. Knowlton, Y. Yao, C. C. Wang and C. P. Hill, Nature, 2000, 408, 115.
50 E. Jankowska, M. Gaczynska, P. A. Osmulski, E. Sikorska, R. Rostankowski, S. Madabhushi, M. Tokmina-Lukaszewska and F. Kasprzykowski, Biopolymers, 2010, 93, 481.
51 J. Witkowska, P. Karpowicz, M. Gaczynska, P. A. Osmulski and E. Jankowska, J. Pept. Sci., 2014, 20, 649.
52 M. Groll, H. Brandstetter, H. D. Bartunik, G. Bourenkow and R. Huber, J. Mol. Biol., 2003, 327, 75.
53 M. Orlowski, C. Cardozo, M. C. Hidalgo and C. Michaud, Biochemistry, 1991, 30, 5999.
54 P. A. Osmulski, M. Hochstrasser and M. Gaczynska, Structure, 2009, 17(8), 1137.
55 G. G. Hammes, Y. C. Chang and T. G. Oas, Proc. Natl. Acad. Sci. U. S. A., 2009, 106(33), 13737.
56 R. Sprangers, X. Li, X. Mao, J. L. Rubinstein, A. D. Schimmer and L. E. Kay, Biochemistry, 2008, 47(26), 6727.
57 M. Gaczynska and P. A. Osmulski, Methods Enzymol., 2005, 398, 425.
58 A. F. Kisselev, T. N. Akopian and A. L. Goldberg, J. Biol. Chem., 1998, 273, 1982.
59 P. Tsvetkov, N. Reuven, C. Prives and Y. Shaul, J. Biol. Chem., 2009, 284, 26234.
60 M. Bajorek, D. Finley and M. H. Glickman, Curr. Biol., 2003, 13(13), 1140.
